Catalyst design and machine learning for thermocatalytic CO2 methanation: A review

CO2 methanation, as a cornerstone technology for a sustainable carbon economy, and its industrial deployment are currently severely constrained by catalyst performance. Although significant progress has been made through optimizing isolated components of the catalyst, the intrinsic complexity of these catalytic systems and their highly non-linear structure-activity relationships make it exceptionally challenging to optimize high-performance catalysts via conventional methods. Therefore, this review argues that future breakthroughs hinge on the synergistic integration of three fields: reaction mechanisms, catalyst design, and machine learning. A deep understanding of reaction mechanisms is the prerequisite for rational catalyst design, and that machine learning (ML) serves as the indispensable engine driving this transformation. To this end, this review systematically summarizes recent advances in active metals, supports, and interfacial engineering, aiming to outline the solid foundation laid by prior research in this field and analyze the inherent limitations of traditional trial-and-error approaches. Subsequently, we systematically introduce machine learning and its innovative applications in catalyst design. Finally, this review outlines key challenges and future research directions in CO2 methanation catalyst design. By integrating these domains, this review aims to lay the groundwork for developing catalysts that combine high performance.